Fishery catch

Author

Ben Williams

Comparing model outputs of fishery catch.

Code
# load ----
library(dplyr)
library(ggplot2)
library(plotly)
theme_set(theme_bw())

# data ----
catch = read.csv(here::here('data', 'admb', 'catch.csv')) %>% 
  rename(admb_2022 = pred) %>% 
  mutate(year = 1961:2022) %>% 
  left_join(data.frame(year = 1961:2022,
                       rtmb_2022 = readRDS(here::here('data', 'rtmb', 'report1.rds'))$catch_pred))

catch %>% 
  tidyr::pivot_longer(c(rtmb_2022, admb_2022), names_to='model') %>% 
ggplot(aes(year, obs)) +
  geom_point(color='gray') +
  expand_limits(y=0) +
  scale_y_continuous(labels = scales::comma) +
  geom_line(aes(y = value, color = model)) +
  scico::scale_color_scico_d(palette = 'roma') +
  xlab('Year') + 
  ylab('Biomass (t)') -> p1

ggplotly(p1)%>% 
  layout(legend = list( x = 0.75, y = .75))
Figure 1. Fishery catch inputs (gray) and model outputs are shown for the 2022 ADMB and RTMB models.
Code
catch %>% 
  mutate(diff = (admb_2022 - rtmb_2022) / admb_2022) %>% 
  ggplot(aes(year, diff)) + 
  geom_point() +
  geom_hline(yintercept = 0, lty=3, alpha = 0.5) +
  scale_y_continuous(labels = scales::percent) +
  xlab('Year') +
  ylab('Percent difference') -> p2

ggplotly(p2) 
Figure 2. Percent difference in model estimates of fishery catch.